FR2992454A1 - System for providing assistance to agronomic decision making, has cooperative servers, where each connection of work station to cooperative servers is duplicated, and data of cooperative servers and workstations are destocked - Google Patents

Abstract

The system has a main server (10) including a data processing unit and a storage unit. A set of cooperative servers (20, 30) is arranged with a data processing unit and a storage unit. Each cooperative server is connected in a permanent manner to the main server. A set of workstations (21, 22, 31-33, 41) of a user includes the data processing unit and the storage unit on which agronomic data of the user is stored. Each connection of the work station to the cooperative servers is duplicated. Data of the cooperative servers and the workstations are destocked.

Description

GENERAL TECHNICAL FIELD The present invention relates to the field of computer-assisted agronomy.

More precisely, it concerns a decision-making aid system in the cultural technical itinerary (optimization of inputs and cultural practices). STATE OF THE ART Agronomy is the totality of the exact, natural, economic and social sciences and techniques that are used in the practice and understanding of agriculture. These techniques have allowed the continuous increase of agricultural yields. For example, farmers have a battery of plant protection products that effectively fight insects, diseases, weeds, and so on. The use of these products, however, must be done in a thoughtful way: Depending on the culture, the time of year, the type of problem, etc., their effectiveness can completely change. It should also be noted that strict regulations regulate agriculture for food security issues. Farmers must henceforth optimize the management of their plots by combining different issues: - agronomic: adjustment of cultivation practices to the greatest extent of the needs of the plant; - environmental: reducing the footprint of agricultural activity; - economic: increased competitiveness through greater efficiency of practices. 30 Parcel management requires a much finer management, which involves the use of computers. Thus appeared decision-making tools. These tools, based on data from years of experience of agronomists, can indicate to the farmer the optimal choices to make in terms of use of products or practices on their land when faced with this or that problem on a particular plot. However, it turns out that the world of agriculture poses specific computer problems. In fact, the models on which the above-mentioned decision-making tools are based are on the one hand constantly evolving, and on the other hand very "precious" because of the time and resources that had to be accumulated to build them. . This results in reluctance of the providers of these tools to leave full access to the farmers (via copies installed on their equipment). Such copies would soon be out of date. It has therefore been proposed an online operation, models and repositories used remaining on servers of the provider. This mode of operation is not suitable for farmers as it is very difficult and uncertain for them to have internet access when they are in the middle of the fields. Being unable to use the tool (or using it with erroneous data) can have disastrous consequences for the farmer. The present invention aims to solve these difficulties by proposing an agronomic decision support system architecture which guarantees farmers a tool which is always up to date with reliable data, available at all times, and which is not do not risk looting the data to the supplier. PRESENTATION OF THE INVENTION The present invention proposes an agronomic decision support system characterized in that it comprises: a master server comprising data processing means and storage means on which are stored a set of calculation methods and reference agronomic data for the implementation of said calculation methods; at least one cooperative server comprising data processing means and storage means, each cooperative server being permanently connected to the master server; at least one workstation of a user comprising data processing means and storage means on which are stored agronomic exploitation data of the user, each workstation being connected at least intermittently to a cooperative server; the data processing means of the master server, the at least one cooperative server and the at least one workstation being configured so that: - at any time are duplicated on the data storage means of each cooperative server said set of calculation methods and reference agronomic data of the master server; each connection of a workstation to a cooperative server is duplicated on the data storage means of the workstation at least a part of said set of calculation methods and reference agronomic data of said server; cooperative, and secondly be duplicated on the data storage means of said cooperative server said agronomic data of operation of the workstation.

According to other advantageous and non-limiting features of the invention: the master server comprises a network interface configured to alternatively enable a direct connection via the Internet with a workstation of a user; the data processing means of the at least one cooperative server and the at least one workstation are further configured so that at each connection of a workstation to a cooperative server are also duplicated on the data storage means of the workstation at least a part of said set of calculation methods and agronomic operating data of the cooperative server from each of the workstations in intermittent connection with said cooperative server; a directional transactional replication mechanism is implemented between the master server and the cooperative servers, and a bidirectional merging mechanism is implemented between each cooperative server and the workstations in intermittent connection with said cooperative server; at least one context is associated with the set of calculation methods, a context being a means of calling a plurality of calculation methods of the set of calculation methods; the user has one or more predetermined contexts, the calculation methods not being executable directly at the level of a workstation without using a context; the said calculation methods are chosen from among the methods making it possible to calculate an efficacy of a phytosanitary treatment, the respect of a regulation, a phytosanitary risk, a need for fertilization, an environmental indicator. BRIEF DESCRIPTION OF THE FIGURES Other features and advantages of the present invention will be apparent from the following description of a preferred embodiment. This description will be given with reference to the accompanying drawings in which: - Figure 1 is a diagram of an agronomic decision support system according to the invention; FIGS. 2a and 2b show two variants of an agronomic decision support system according to the invention; FIG. 3 is a schematic diagram of a method for evaluating a risk according to the invention; FIG. 4 is a map of France used in one embodiment of a method for evaluating a risk according to the invention; FIGS. 5 and 6 are views of an interface used in an embodiment of a method for evaluating the effectiveness of a phytosanitary treatment according to the invention.

DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT Architecture With reference to FIG. 1, the present invention relates, according to a first aspect, to a system 1 for assisting agronomic decision-making. By "agronomic" we mean all the issues related to agriculture, whether phytosanitary (choice of a program, calculation of a health risk, etc.) or regulatory, economic, etc. . This system 1 has three "levels". At the first level, there is a main server called the master server 10 (which can possibly be composed of several servers). Disposed in particular in the premises of the supplier of the decision-making tool, it comprises data processing means and storage means on which are stored a set of calculation methods and agronomic reference data for the implementing said calculation methods. Said agronomic reference data form a base called "reference frames" of a large volume. This is the data accumulated over years that make the value of a decision support tool.

The standards include in particular data relating to active ingredients (i.e., molecules having phytosanitary effects), products, phytosanitary treatment programs, uses, types of crops, plot configurations, etc. This is all data allowing the setting of all the said calculation methods, which is why we also speak of "common" data repositories. These methods are the functions whose execution results in the resolution of the cases for which decision support is required, depending on the reference data and farm data (which will be described later) of the farmer using the decision support tool. Calculation methods make it possible, for example, to calculate the effectiveness of a phytosanitary treatment, the respect of a regulation, a phytosanitary risk, a need for fertilization, an environmental indicator, etc. Some of these methods will be described in detail later. The calculation methods and the reference data can for example be modified at this master server 10 only by means of a specific and secure interface (for example a single workstation), also arranged in premises of the provider. Combining all relevant reference data with all computational methods in a single database, and combining a wide variety of computational methods in a single set, also minimizes the number of interventions to be performed at a single level. of the master server 10 to perform an update. In addition, it reduces the overall volume of data (and therefore the bandwidth required for updates), since this avoids any "duplicate" data. Finally, it also allows to easily add a new method without touching to give this reference. This "unified" structure of the decision support tools provided by the system according to the invention is therefore very advantageous. The second level, visible in FIG. 1, consists of at least one cooperative server 20, 30. By cooperative is meant a group of users who combine to pool resources, be it hardware , premises (storage spaces for equipment) or data. By "cooperative server" is meant any secondary server (compared to the master server 10) located at the local level and bringing together a plurality of professionals to work together. On average, there are about 50 users at the level of a cooperative. The cooperative servers comprise data processing means and storage means, each cooperative server 20 being permanently connected to the master server 10. These cooperative servers 20, 30 are in fact fixed and have means of permanent network connection, for example via Ethernet. The connection between the master server 10 and the cooperative servers 20, 30 is in particular through the Internet. By "permanent" is meant substantially discontinuous. It will thus be understood that this connection may be interrupted from time to time briefly, for example to carry out maintenance or in the event of a mains failure, but most of the time the connection is active. Advantageously, each of these cooperative servers 20, 30 is isolated from the rest of the Internet network by a firewall ("firewall" in English terminology), configured to block any traffic that is not intended for data replication. agronomic, as will be shown later. The third and last level consists of at least one workstation 21, 22, 31, 32, 33, 41 of a user. Each workstation depends on a single cooperative server 20, 30. (With one exception, represented by the station 41, which will be described later). For example, in FIG. 1 the stations 21 and 22 depend on the cooperative server 20, and the stations 31, 32 and 33 depend on the cooperative server 30. Each workstation 21, 22, 31, 32, 33 is connected to its cooperative server 20, 30 at least intermittently, preferably still via the Internet. "Intermittent" is opposed to "permanent" in that system 1 allows a workstation 21, 22, 31, 32, 33, 41 to be disconnected several hours a day or more. As explained previously, it is indeed common for the user to be in the middle of a field, out of reach of any Wi-Fi network via which his workstation 21, 22, 31, 32, 33, 41 could establish. a connection with its cooperative server 20, 30. However, it will be understood that it is not limited to a discontinuous mode, and it is quite possible that one or more workstations 21, 22, 31, 32 , 33, 41 are connected to their cooperative server 20, permanently. The at least one workstation 21, 22, 31, 32, 33, 41 can be any kind of computer equipment, including a mobile terminal (smartphone, touch pad, etc.). It comprises data processing means and storage means on which are stored agronomic operating data of the user. This operating data is the personal data of the user. They describe its plots, crops, stage of development, treatments made and / or planned, etc. This data can be entered manually by the user, for example by input means of the workstation 21, 22, 31, 32, 33, 41 of the user such as a keyboard or a touch interface, or via specific equipment arranged on its plots and connected (possibly intermittently) to the workstation, which directly generate operating data. On each level, the data is advantageously stored in the form of one or more relational databases, the operations on these bases being in particular carried out using the SQL ("Structured Query Language") language. In fact, coupled with replication mechanisms, the three-level architecture allows, as we will see later, optimal data propagation between the different actors for a perfect availability of decision support tools and security. maximum data.

Mechanisms of replication The system 1 according to the invention is remarkable in that it implements particular data replication mechanisms in the structure described above. Indeed, the updates of the set of calculation methods and / or reference data are done at the level of the master server 10 by the provider of the decision support tool (via the possible secure interface previously described). These replication mechanisms allow the propagation of these updates to the workstations 21, 22, 31, 32, 33, 41 without the disadvantages of the prior art mentioned above. The data processing means of the master server 10, the at least one cooperative server 20, 30 and the at least one workstation 21, 22, 31, 32, 33, 41 are in particular configured so that to implement two replication mechanisms, preferably via SQL requests as explained. The first mechanism aims at replication between the first two levels (master server 10 and cooperative servers 20, 30), and the second aims at replication between the second and third levels (cooperative servers 20, 30 and workstations). 21, 22, 31, 32, 33, 41). The three-level structure means that at any time the master server 10 needs to communicate only with the cooperative servers 20, 30, which are much less numerous than the workstations 21, 22, 31, 32. , 33, 41. It would indeed be impossible in terms of bandwidth (coupled in addition to the problem of intermittent connection of workstations) to be able to update at any time all users. The architecture of the system 1 thus allows a very good availability of the data, and facilitates the updates. The first mechanism is advantageously a so-called "unidirectional transactional replication" mechanism. In a general manner, this first mechanism ensures that, at any given instant, the set of calculation methods and agronomic data of the master server 10 are duplicated on the data storage means of each cooperative server 20. In other words, any update of the data at the master server level automatically results in the same instant update at each cooperative server 30 20, 30. This is made possible by the fact that the connection between these servers is permed. This replication mechanism is unidirectional, which means that any modification of the reference data at the level of a cooperative server 20, 30 which could take place voluntarily or otherwise (for example deletion of part of the data, attempt to replace method, etc.) not only does not propagate to the master server 10, but is immediately corrected by reduplication of the master server data. In addition to a possible firewall, the second level serves as a "screen" that protects the master server from any risk of data corruption. In the advantageous "transactional" mode, the replication (for example of an update of some of the reference data at the master server level), then called a transaction, which makes the database go through (here the copy of the reference data at a cooperative server 20, 30) of a state A prior to the transaction at a later state B, is implemented via a series of operations respecting atomicity, coherence, insulation and durability (criteria abbreviated by ACID). The transaction is atomic: the sequence of operations is indivisible, in case of failure during one of the operations, the sequence of operations must be completely canceled (SQL ROLLBACK command) whatever the number of operations already successful.

The transaction is consistent: the content of the database at the end of the transaction must be consistent without each transaction during the transaction gives consistent content. Inconsistent final content must result in the failure and cancellation of all transactions in the transaction.

The transaction is isolated: when two transactions A and B are executed at the same time, the modifications made by A are not visible by B, nor modifiable by B until transaction A is completed and validated (SQL COMMIT command) . The transaction is sustainable: once validated, the state of the data base must be permanent, and no technical incident such as a connection break should be able to cause a cancellation of the operations performed during the transaction.

The respect of these criteria guarantees to the user that at all times the data storage means of each cooperative server 20, 30 comprise an exact and perfectly up-to-date copy of the set of calculation methods and the agronomic data of reference of the master server 10. The second replication mechanism is advantageously a so-called "bidirectional fusion" mechanism. In general, this mechanism ensures that at a connection of a workstation 21, 22, 31, 32, 33, 41 to a cooperative server 20, on the one hand is duplicated on the data storage means of the workstation 21, 22, 31, 32, 33, 41 at least part of said set of calculation methods and reference agronomic data of said cooperative server 20, 30, and secondly be duplicated on the storage means of data of said cooperative server 20, said agronomic data of operation of the workstation 21, 22, 31, 32, 33, 41. It is noted that this replication mechanism is implemented during a connection. Indeed, in the case of an intermittent connection, the replication mechanism can be started each time a connection is established between a workstation 21, 22, 31, 32, 33, 41 and its cooperative server 20 30. The system according to the invention is however not limited to this regularity, and it will be understood that the frequency of implementation of the replication depends on the periods of time during which a workstation is connected or not.

For example, in a case where the workstation 21, 22, 31, 32, 33, 41 is permanently connected, one can imagine either to implement at any moment a transactional replication similar to that of the first mechanism, or that the replication is attempted at regular intervals (eg every night at midnight).

Alternatively, in the case of an intermittent connection, it is possible to implement a timer (for example 24 hours), the replication being implemented either at the expiration of the timer if the workstation 21, 22, 31, 32, 33, 41 is connected, either to the first connection after the expiration of the timer. Those skilled in the art will be able to adapt the invention on a case by case basis. In any case, it should be noted that the replication is bidirectional, that is to say that not only data is copied from the cooperative server 20, 30 to the workstation 21, 22, 31, 32, 33 , 41 (downward direction), but also in the other direction (upward direction). The data that is copied in the downward direction is all or part of the set of cooperative methodologies and agronomic data of the cooperative server 20, 30 (which are an exact copy of those of the master server 10). The transfer can be partial, if for example the user has subscribed to only some of the methods proposed by the decision support tool. It is therefore unnecessary to consume bandwidth to transfer data that it will not use.

It also allows for increased data security. The data that is copied in the uplink direction is the user's agronomic data stored on his workstation 21, 22, 31, 32, 33, 41. As explained previously, this is the personal data of the user. user, on the basis of which the calculation methods from the master server 10 can be executed. The feedback of user operating data means that each cooperative server 20, 30 retrieves the data of each of its members, as can be seen in FIG. 2a, which makes it possible to have an overall view of of the cooperative, and to implement a common strategy. According to an alternative embodiment represented by FIG. 2b, the data processing means of the at least one cooperative server 20, 30 and the at least one workstation 21, 22, 31, 32, 33, 41 are further configured so that at each connection of a workstation 21, 22, 31, 32, 33, 41 to a cooperative server 20, are also duplicated on the data storage means of the workstation. Work 21, 22, 31, 32, 33, 41 at least a part of said set of calculation methods and agronomic data of operation of the cooperative server 20, 30 from each of the workstations 21, 22, 31, 32, 33, 41 in intermittent connection with said co-operative server 20, 30. In other words, the bidirectional merge is "total", which means that a user's operating data is reassembled to a co-operative server 20 , 30 are redistributed to all workstations 21, 22, 31, 32, 33, 41 connected to the cooperative server 20, 30, along with the set of calculation methods and the reference agronomic data. Thus, each user has the operating data of all the users of his cooperative, which enables him to implement the decision support tools taking into account the parameters of his co-subscribers, in order to favor even more the emergence of a common strategy. In this so-called non-partitioned mode, we also note that subscriber stations do not need to be nominative (they share the same 15 data in the end). This further simplifies management. It should be noted that the choice between these two modes of operation can be made cooperative by cooperative. Web Services and Contexts Mode Advantageously, there are also options for system 1 according to the invention. As seen in FIG. 1, the master server 10 may comprise a network interface configured to alternatively enable a direct connection via the Internet with a user's workstation 21, 22, 31, 32, 33, 41. This is to allow calculations via Web Services, alternatively to local calculations on the workstation 21, 22, 31, 32, 33, 41 with the data propagated by replication. In this mode, a browser calls a function by sending a stream eg XML (in the "eXtensible Markup Language" markup language) from a user interface, for example a workstation browser. This flow contains the data (from the operating data of the user) necessary for the implementation of the calculation methods. The XML stream may optionally be encapsulated in an encryption layer such as SSL ("Secure Sockets Layer"). The processing means that then carry out the execution of the selected functions are then those of the master server 10 and not those of a workstation 21, 22, 31, 32, 33, 41 It may be the same functions of decision-making than the main mode of use, but for example via a less high-end subscription (so for other users than those benefiting from replication, for example users not linked to a cooperative: the case shown in Figure 1, where the workstation 41 does not depend on any cooperative server). This operating mode requires the internet connection, and does not allow the recovery of the set of calculation methods and reference data on the workstation 41 of the user. Alternatively, this alternative mode of operation is open to all those having the local mode of operation, but reserved for certain functions, for example those for which the provider does not wish to communicate reference data. Still alternatively, this mode is provided for workstations 21, 22, 31, 32, 33, 41 of a different type, for example smartphones, which have easier access to the Internet (via 3G), but lower resources both in terms of processing power and storage space. In any case, since the use of the system 1 is essentially local, this Web Services mode increases the bandwidth requirements of the system very little. Moreover, as we always see in Figure 1, the set of calculation methods can be associated with predetermined "contexts", each user having all or part of these contexts. A context is an ordered sequence of methods of the set of computation methods to be executed totally or partially. In other words, a context can be seen as a script whose execution requires a number of methods. A context is modeled by an XML file. For example, the methods can be grouped by category (for example, use, approval, mixing, etc.), and a context 5 allows, as desired, to execute only the methods of a category or to combine several (use + mixing, use + mixing + frequency, etc.). Contexts offer flexibility to automate decision support tools and in particular make it possible to set up a few specificities depending on a region, a country or conditions of use of the service 10 depending on whether to control a product, a solution or all the solutions of a parcel. In addition, the contexts facilitate possible updates of the functions. Finally, more specifically, the contexts can make it possible to increase the security of the data. The calculation methods are thus advantageously not executable directly at a workstation 21, 22, 31, 32, 33, 41 of a user without using a context. This allows for example to restrict the use of methods based on subscription categories, or to prevent unscrupulous users from "reverse engineering" on calculation methods and reference data. Will now be described innovative calculation methods that can offer the system according to the invention. In general, the 25 calculation methods are chosen from methods making it possible to calculate an effectiveness of a phytosanitary treatment, the respect of a regulation, a phytosanitary risk, a need for fertilization, an environmental indicator. Risk Calculation According to a second aspect, the invention thus relates to a method for processing agronomic and geoclimatic data of a crop of a parcel stored on storage means of a workstation 21, 22, 31, 32 , 33, 41 for the evaluation of a risk related to a pathology on the culture, said pathology being chosen from a plurality of reference pathologies.

This method is implemented by means of processing the workstation 21, 22, 31, 32, 33, 41. It will however be understood, by way of exception, with reference to the particular embodiment of the aid system 1. in the decision-making according to the first aspect of the invention in which Web Services is used that the method can be implemented, at least partially, at the level of the master server 10. The purpose of this method of calculation is to dynamically assess the risks to pathologies of crops, in particular straw cereals (that is to say in particular cryptogamic diseases causing damage to the crop and loss of yields). By culture is meant a cultivated plant species, and by reference pathologies is meant a plurality of pathologies that have a similar impact and similar exposure. The invention is not limited to any particular set of reference pathologies. Advantageously, it is Fusarium wilt, septoria, and brown rust. Alternatively, as we will see septoria and brown rust are very similar, and we observe particularly good results by taking into account only these two pathologies in the set of reference. Fusarium wilt affects grains that look rough and mangy, and have a light brown, pink, or white-gray color.

This pathology is caused by certain Fusarium fungi commonly present in soils, but having in these cases a parasitic development. There are indeed many species of Fusarium, some of which are only pathogens and / or are likely to emit mycotoxins (fusariotoxins in this case).

Septoria and brown rust are so-called foliar pathologies, that is to say that they cause lesions of the leaves. Septoria and brown rust are also fungal diseases of plants (mainly caused by Septoria tritici and Septoria nodorum for the first, and Puccinia recondita for the second). They are particularly common in wheat and other grains and are found in all growing areas around the world. They can cause yield losses of more than 40%. Epidemics can be very damaging because they are epidemics with exponential development. A risk assessment is an estimate of the probability of being affected by the condition, and possible performance losses in this case. As we will see, the process allows the risk to be compared in various treatment scenarios, which is of great help in making the decision to treat (treatment, if, yes, what treatment, at what stage of development of the culture? etc.). Before starting the process, is advantageously implemented a culture eligibility test. Indeed, as explained above, some cultures are not sensitive to this or that pathology. It is then useless to continue the calculation. In the case where the cultivation of the plot studied is eligible for the calculation of the risk, the process for the evaluation of a risk related to the pathology takes into account agronomic and geoclimatic parameters. This is a subset of the reference data and operating data mentioned above. In particular, as can be seen in Figure 3 which schematically represents the process, the agronomic and geoclimatic parameters include a sowing factor y, a variety factor 6, a climatic factor a and a regional factor p. Each of these factors is here a numerical value around 1. The lower one of these factors, the more likely the culture is to undergo a pathology. It is important to note that these parameters are not specific to the pathology being studied. The distinction between the pathologies of the reference disease group is further downstream in the method.

The sowing factor is the first agronomic parameter. I l corresponds to the effect of sowing date. Indeed, we note for example that late crops are more resistant. The variety factor 6 is the second agronomic parameter. It corresponds to the predisposition of a variety to the pathologies of the reference set. Among the eligible crops, some are naturally quite resistant, while others are sensitive. In general, many varieties are known to have a Global Sensitivity Score (NSG) on a scale of 1 to 10.

Advantageously, the calculation of the variety factor is given by the formula 6 = 0.95 + 0.02 * (NSG -1). The climatic factor a is the first geoclimatic parameter. It corresponds to the effect of climate. Classes of values are given among the reference data as a function of climatic episodes (the humidity favors for example the disease). Finally, the regional factor 13 is the second geoclimatic factor. By default, a value is given by department. For example, Figure 4 shows value classes by department on a French map. It will be appreciated that the above examples are merely illustrative and that the invention is in no way limited to a particular set of agronomic and geoclimatic parameters. With these parameters, the processing means of the workstation 21, 22, 31, 32, 33, 41 of the user calculate a parameter called Global Inability (NG). The parameter NG, expressed as a percentage, represents the risk that the parcel contracts a pathology, but is not yet specific to said pathology studied, as explained above. By way of example, the parameter NG can be given by the formula NG = 1.65 * (1-a * 13 * y * δ). In the case of septoria and brown rust, the NG parameter is more precisely a parameter of Global Foliar Nuisance (NGF).

In a particularly advantageous manner, the calculation of the parameter NG is carried out "dynamically". This means that the process is automatically restarted as soon as the value of a parameter is changed. This allows the value of some parameters to be initially absent or ignored. For this purpose, the reference data set includes a default value for each of the agronomic and geoclimatic parameters. The workstation processing means 21, 22, 31, 32, 33, 41 initially attempt to obtain a value (for example through the processes explained above), and in case of failure use the default value. In case of modification of the user's operating parameters and / or update of the reference database, an attempt is made to obtain the missing parameters. The process is then dynamically restarted, in a manner totally transparent to the user if at least one of the missing values is obtained. This is repeated iteratively. In a second step, the processing means of the workstation 21, 22, 31, 32, 33, 41 of the user calculate a specific risk of harm rate of said pathology according to the parameter NG and sensitivity parameters 20 culture with respect to said plurality of reference pathologies. In particular, the rate of specific harm N of the pathology P is advantageously given by the formula (10-NS (P)) * weight (P) N (P) NG * the pi being the different (10-NS ( p,)) * weight (p1) '25 pathologies, NS (p) being the varietal sensitivity note to pathology p, and weight (p) being the ratio of pathology p compared to pi pathologies observed locally. The sigma means a sum on the pathologies of the set of reference. Varietal sensitivity notes are parameters derived from the reference data. The idea is to isolate the harm caused by the specific pathology under study. The weights of the different pathologies are indeed values corresponding to the occurrence of such pathology compared to others observed locally, chosen such as weight (p1) = 1. For example, it is observed that in the north of France the brown rust is four times more present than the septoria, whereas it is the opposite in the south.

In a third step, the processing means make it possible to express a risk of damaging the said pathology in the form of an expected yield loss as a function of the said level of risk of damage and yield objectives of the parcel. In particular, this is achieved for example by multiplying the rate of specific harm risk obtained by a yield objective expressed in quintals per hectare. As explained previously, this value can be left empty, and a default value is temporarily used (for example, for soft wheat, the default target is 80 q / ha). The value obtained is the expected loss of return. Indeed, if the risk of specific harm is zero (no risk that the culture is affected by the pathology) then the expected loss is 0 q / ha. In a fourth step, also visible in Figure 3, the data processing means take into account phytosanitary treatments made or planned. The value of expected yield loss can thus be corrected by calculating a possible reduction of the nuisance according to possible phytosanitary treatments received by the parcel. The risk of harm after reduction obtained at the end of this step is advantageously equal to the expected yield loss 25 multiplied by a coefficient inversely proportional to the effectiveness of the phytosanitary treatment or treatments received. In particular, in the absence of treatment, this coefficient is equal to 1. The expected loss of return does not change. Advantageously, this step takes into account the durations of protection offered by a treatment. Indeed, a phytosanitary treatment will initially prevent the appearance of a pathology (or even stop it if it is not a preventive treatment), but this protection will gradually fade. The risk therefore falls initially, then goes back up. By way of example, a coefficient of effectiveness of a treatment against DP seporosis treatment can be given by the formula c = DP treatment 0.55 * N (P) + 13.12 being the duration of protection offered by the treatment implemented. If the user enters that he has carried out a treatment but does not specify the duration of protection offered by this treatment, the default value taken by the processing means for the coefficient is for example 0.3.

In a final step, a display on interface means (such as a screen) of said workstation 21, 22, 31, 32, 33, 41 of a risk level related to said pathology on the culture according to the results obtained is implemented. Advantageously, the risk is interpreted according to four classes - if the expected yield loss after reduction is less than 4 q / ha, the level of risk is low; - if the expected yield loss after reduction is between 4 and 10 q / ha, the level of risk is medium; - if the expected yield loss after reduction is between 10 and 25 q / ha, the level of risk is high; - If the expected yield loss after reduction is greater than 25 q / ha, the risk level is very high. A color code on the user interface advantageously makes it possible to visualize these levels of risk in a particularly ergonomic manner. Given the dynamic mode, this risk is recalculated continuously. At any time, taking into account a new parameter can make a significant change in the risk, and cause a change of level. This is immediately visible on the user interface. It is also possible to provide workstation alerts 21, 22, 31, 32, 33, 41 automatically triggered in the event of a change in a level of risk, so that the user is immediately warned to apply at most quickly the appropriate measures. On the other hand, it should be noted that other parameters, such as whether or not plowing has been done, a previous type of crop, etc. can be taken on a case by case basis to refine the results relating to a pathology. Efficiency of treatments According to a third aspect, the invention furthermore relates to a method of processing agronomic and geoclimatic data of a crop of a parcel stored on storage means of a workstation 21, 22, 31, 32, 33, 41 for evaluating the efficacy of a phytosanitary treatment against a given target (see below), said phytosanitary treatment comprising the application of at least one phytosanitary product comprising at least one active principle to a given dose. This method is also implemented by means of processing the workstation 21, 22, 31, 32, 33, 41. However, it will again be understood, by way of exception, with reference to the particular embodiment of the system. decision support according to the first aspect of the invention in which Web Services is used, that the method can, at least partially, be implemented at the level of the master server 10. The purpose of this method of The purpose of the calculation is to evaluate the effectiveness of a phytosanitary treatment comprising the application of one or more products to combat a given phytosanitary problem. By phytosanitary problem will be understood any affection of the culture caused by the presence of organisms (cryptogamous, pests, adventitious) which can be harmful to him and negatively impact its yield. Thus, the term "target" is more particularly used to designate the pest that can be targeted by a phytosanitary treatment. It should thus be noted that, unlike the method for the evaluation of a risk described above, which was adapted only to pathologies, that is to say diseases of the culture caused by a disease, the evaluation method of 'an effectiveness is also suitable for targets such as pests (insects, etc.) and weeds (in other words weeds). According to two particularly advantageous alternative methods, the method concerns on the one hand the evaluation of the effectiveness of a fungicidal phytosanitary treatment, and on the other hand the evaluation of the effectiveness of an herbicide phytosanitary treatment. These two embodiments will be described later.

A phytosanitary treatment corresponds to the application of one or more products at a given dose, that is to say a volume of product per hectare. This product contains one or more active ingredients (or "active ingredients") at a fixed concentration. By way of example, a product A comprises two active ingredients, namely prothioconazole at 150 g / l and bifaxen at 75 g / l. These are antifungals. Treatment with this product alone at 0.8 L / ha results in 120 g / ha of prothioconazole and 60 g / ha of bifaxen. As will be seen later, it is possible to provide several passages at different times in a treatment. The efficiency calculation method according to the invention begins with a selection by the user, thanks to an input interface of his workstation 21, 22, 31, 32, 33, 41, a range of stages of development of the culture during which the application of said phytosanitary treatment is provided. By "stage of development" is meant a visually identifiable stage in the growth of a crop. For example, for wheat, there are 44 stages of development such as "1 node", "appearance of the last leaf", "heading", "early flowering", "senescence", etc. Some of these stages are visible in Figure 5. Through its interface, the user selects his range by choosing a start of range stage and an end of range stage. In Figure 5, two ranges of different extents are shown.

It should be noted that the range may be discontinuous (in other words a union of several continuous ranges). In a first treatment step, the workstation processing means 21, 22, 31, 32, 33, 41 calculate for each active principle and each stage of development of the culture of said range a specific rate of effectiveness. of the active ingredient at the given dose against said target at this stage of development. It is understood that the larger the selected range, the longer the calculation.

A specific effective level EFF (P) of an active ingredient P is obtained at a dose against said target at a development stage. The calculation of this specific efficiency rate differs according to the type of target. In the case of a fungicidal target (fungicidal treatment), there is no particular formula, effective levels of an active ingredient at one stage being known as soon as a threshold value of dose. The step then consists of a comparison of the dose given with the threshold dose in each case and the interrogation of the reference data. In the case of an adventitious target (herbicidal treatment), a specific level of efficacy EFF (P) of an active ingredient P at a dose against said target at a stage of development is given by way of example by the formula EFF (P) = EFF * (1 1) EFF, ', d50 and b being 1+ (d / d50) b reference parameters depending on the active principle, the culture, the target and the stage of development . d50 is a threshold dose which corresponds to a dose where the efficacy is equal to half of the maximum efficacy EFF, '. By way of example with reference to FIG. 6, the case of a product B (which is an antidicolytedone whose only active ingredient is prosulfocarb) at a dose of 4.5 L / ha on soft wheat against the fumeterre (a wild plant) gives an efficiency of 87,01% at the "young" stage and 58,18% at the "developed" stage. In a second step, the method combines the active ingredients so as to calculate for each stage of development of the culture of said range an overall rate of effectiveness of treatment against said target at this stage of development, depending on the levels of effectiveness of the active ingredients of the phytosanitary products of the phytosanitary treatment at this stage of development.

A global effectiveness rate EFF of a phytosanitary treatment against a target as a function of the effectiveness rates of the active ingredients Pi of phytosanitary products of the phytosanitary treatment at this stage of development is advantageously obtained either via the formula EFF = 1_11 (1 - EFF (Pi)) (this is a geometric mean), either by the formula EFF = 1 (EFF (Pi)), K being a coefficient dependent on the target (it IK is then an average weighted arithmetic). The first case is preferred in the case of a herbicide treatment and the second is preferred in the case of a fungicidal treatment. In the latter case, the coefficient K is fixed. For example, if the target 15 is trampling or powdery mildew, it is 1. For the brown rust mentioned above, more resistant, it is 1.3. In any case, the specific efficacy levels of an active ingredient at a dose against said target at a developmental stage can be weighted by resistance coefficients. In the absence of such coefficients, default values are used. As explained above, the treatment may include several passages, possibly at different growth stages of the culture. In this case, the overall rate of effectiveness is calculated for each of the passages, and the total efficiency is the sum of the overall efficiencies of the passages. If in one or the other of the calculations one exceeds 100%, 100% is kept as final value. The means of treatment finally implement a comparison of the calculated global efficiency rates so as to identify an optimum development stage where the overall effectiveness of the phytosanitary treatment against said target is the highest. In the case of a multi-pass program, this step is repeated passage-by-passage. In a last step, as in the risk calculation method, a display on interface means of said work station 21, 22, 31, 32, 33, 41 of a level of phytosanitary treatment efficiency on the target of after the results obtained allows is implemented. This process thus makes it possible, by consuming few resources, to directly obtain the optimal combination in terms of treatment planning. The user then knows the maximum efficiency he can get, and knows immediately if this treatment is worth it. It can also easily compare a plurality of products, and act optimally. Advantageously, the interpretation of the efficiency is done according to 5 classes: - if the overall efficiency at the optimal stage is less than 50%, the level of efficiency is insufficient; - if the overall efficiency at the optimal stage is between 50 and 70%, the efficiency level is low; - if the overall efficiency at the optimal stage is between 70 and 90%, the level of efficiency is good; - if the overall efficiency at the optimal stage is between 90 and 100%, the level of efficiency is very good; - if the overall efficiency at the optimal stage reaches 100%, the level of efficiency is excellent.

A color code on the user interface advantageously makes it possible to visualize these levels of efficiency in a particularly ergonomic way. Example An example of phytosanitary treatment (on wheat, against brown rust) in which two passes are made, the first involving the application of a product C at 1.33 L / ha, and the second the application of product D at 1 L / ha and product E at 0.3 L / ha. The active ingredients of the first pass are propiconazole, cyproconazole and chlorothalonil. The active ingredients of the second pass are boscalid, epoxiconazole and pyraclostrobin.

The first pass is at the "2 knots" stage and the second between the "Last leaf" and "Flowering" stages. The implementation of the method makes it possible to obtain the following results: 1st pass (the optimum stage is "2 knots") EFFpropiconazole = 3% EFFcyproconazole = 20% EFFchlorothalonil = 2% 2nd pass (the optimum stage is "Apparition of the last leaf) EFFboscalid = 0% EFFepoxiconazole = 50% EFFpyraclostrobin = 35% The coefficient K being 1.3 for brown rust, we obtain an overall efficiency of the first pass of 19% and an overall efficiency of the second pass of 65%. The total efficiency rate is thus 84%, a "good" level.

Systems According to a fourth and fifth aspect, the invention relates to systems for implementing the methods object of the second and third aspects of the invention. These systems can be much more modest than the system according to the first aspect of the invention. In particular, these systems 1 comprise at least one work station 21, 22, 31, 32, 33, 41 comprising data storage means and data processing means configured for implementing the data processing method. Agronomic and geoclimatic characteristics of a crop of a plot for the evaluation of a risk related to a pathology on the crop according to the second aspect of the invention and / or the method of processing agronomic and geoclimatic data of a crop a plot for evaluating the effectiveness of a phytosanitary treatment against a given target according to the third aspect of the invention. In a particularly preferred manner, system 1 according to the first aspect is used for these methods, which, as previously explained, makes it possible, by virtue of its common data standards, to implement these two methods (and others) by itself, but invention is not limited to him.

Claims (7)

REVENDICATIONS1. System (1) for agronomic decision making, characterized in that it comprises: a master server (10) comprising data processing means and storage means on which a set of calculation methods are stored; and agronomic reference data for the implementation of said calculation methods; at least one cooperative server (20, 30) comprising data processing means and storage means, each cooperative server (20, 30) being permanently connected to the master server (10); at least one workstation (21, 22, 31, 32, 33, 41) of a user comprising data processing means and storage means on which the user's agronomic data are stored; each workstation (21, 22, 31, 32, 33, 41) being at least intermittently connected to a cooperative server (20, 30); the data processing means of the master server, the at least one cooperative server (20, 30) and the at least one work station (21, 22, 31, 32, 33, 41) being configured from so that: - at any time are duplicated on the data storage means of each cooperative server (20, 30) said set of calculation methods and reference agronomic data of the master server (10); at each connection of a workstation (21, 22, 31, 32, 33, 41) to a cooperative server (20, 30) on the one hand is duplicated on the data storage means of the workstation (21, 22, 31, 32, 33, 41) at least a part of said set of calculation methods and reference agronomic data of said cooperative server (20, 30), and secondly be duplicated on the means of destocking of data from said cooperative server (20, 30) said agronomic operating data of the workstation (21, 22, 31, 32, 33, 41). 2. System according to claim 1, wherein the master server (10) comprises a network interface configured to alternatively enable a direct connection via the Internet with a workstation (21, 22, 31, 32, 33, 41) of a user. 3. System according to one of the preceding claims, wherein the data processing means of the at least one cooperative server (20, 30) and the at least one workstation (21, 22, 31, 32, 33, 41) are further configured so that at each connection of a workstation (21, 22, 31, 32, 33, 41) to a cooperative server (20, 30) are also duplicated on the data storage means of the workstation (21, 22, 31, 32, 33, 41) at least a part of said set of calculation methods and agronomic data of operation of the cooperative server (20, 30) originating from each of the workstations (21, 22, 31, 32, 33, 41) in intermittent connection with said cooperative server (20, 30). The system of one of the preceding claims, wherein a directional transactional replication mechanism is implemented between the master server (10) and the cooperating servers (20, 30), and a bidirectional merging mechanism is implemented. between each cooperative server (20, 30) and the workstations (21, 22, 31, 32, 33, 41) in intermittent connection with said cooperative server (20, 30). 5. System according to one of the preceding claims, wherein at least one context is associated with the set of calculation methods, a context being a means for calling a plurality of calculation methods of the set of methods of calculation. calculation. 6. System according to claim 5, wherein the user 5 has one or more predetermined contexts, the calculation methods not being directly executable at a workstation (21, 22, 31, 32, 33, 41) without using context. 7. System according to one of the preceding claims, wherein said calculation methods are chosen from among methods making it possible to calculate an efficacy of a phytosanitary treatment, the respect of a regulation, a phytosanitary risk, a need for fertilization, an environmental indicator.